se, for example, J.R. Smith, in Interactions on Metal Surfces edited by R. Gomer (Springer-Verlag, New York, 1979) >I, R. Lapidus, Am. J. Phys 37, 9301969} 37, 1064(1969}; 38, 905 (1970); 442, 316(1974} 43, 790(1975} 46,1281 (1978), $0,453 (1982), $0, $62 (1982), '50, 563 (1982) 50, 663 (1982); B. James, ibid. 38, 13191970}; LI. Folly, ‘id. 44,1192 1976}, 46,889 (1977); M. K. Srvistava and R. K. Bhaduri, bid. 45, 462 (1977) [also see M. K. Srivstava, R. K. Bhaduri, and A. K. Datta, Phys. Rev, A14, 1961 (1976): L. D. Nielson, Am. J, Phys 46, 889 (1973), V. Uromow, ibid 47, 278 1979} 8. T. Epstein, bid. 28, 495 1960}, E. Kujawski, bid. 99, 1248 (1971), W.C. Damen, ibid 43, $31 (1975) D. ‘A. Atkinson and H. W. Crater, ibid. 43,301 (1975), C. U-Segee and J.D. Sullivan, ibid. 44, 729 (1976); A. Sauceda and W. van Dijk, bid 46, 1195 (1978) 8. A. Mosekowski and G. Strobel, tid. 47, 943 (1979). “There are numerous publications in theliterature which use the &-function ‘potential to investigate the properties of many-electron atoms, Bose-Ein- ll guide the reader to many other papers: A. A. Frost, J.Chem Phys. 22, 1613 (1954) 25, 1150 1956) 25, 11541956); E. Lieb and W. Linigr, Phys, Rev. 130, 1605 (1963; 190, 1616 1963}, J. M. McGuire, J. Math. Phys. 5, 622 1964); 6,432 (1965}, 7,123 (1966), C. N. Yang, Phys. Rev. Lett. 19, 1312 (1967 Phys. Rev. 168, 1920 (1968), C-N. Yang and C. P. Yang, J Math. Phys 10, 115 1969 C.P. Yang, Phys. Rev. A2, 1541970} C.K. Laiand C.N. Yang, iid-3, 393 (1971),C. K. Lai ibid. 8, 2567 1973), CK. Lai, J. Math Phys. 18, 954 (1974) E. Lieb and M. de Llano, iid. 19,860 (1978) E. M. Haacke and L. Foy, Phys. Rev. C23, 1320 1981}, 23, 1330, (198, Differential forms as a basis for vector analysis—with applications to electrodynamics Nathan Schieifer® Department of Phsis, Temple University, Philadelphia, Pennsylvania 19122 (Received 26 March 1982; accepted for publi mn 25 August 1982) 1 give a concise and self-contained presentation of the theory of differential forms and how it subsumes all of classical vector analysis when applied to a three-dimensional space. The differential form analog of all vector operators, identities, and theorems are given (many of which are proved), all in the context of classical electrodynamics. I. INTRODUCTION Vector analysis is usually introduced to physicists in the context of classical electrodynamics. The theorems of Stokes and Gauss allow one to convert integral equations involving the E and B fields into differential equations— the Maxwell equations. The similarity between the two theorems is striking, in that each of them states that the integral of some type of derivative of a vector field over @ region of certain dimension is equal to the integral of the vector field over the boundary of that region. And yet there does not seem to be a deeper underlying structure connect- ing the two theorems. Search as one may, classical vector analysis does not supply any answers. ‘Though this conceptual coincidence may be the most compelling one, itis by no means unique. Take, for exam- ple, the fact that curl-grad = 0 and div-curl = 0. The for- mer equation applies of course to a scalar field and the latter to a vector field. Could there possibly exist a more general type of “derivative” operator that would give us both these equations? Moreover, could not this new type of derivative shed light on the connection between Gauss's and Stokes’s theorems? AA deeper and unifying structure does indeed exist, and is provided by the theory of differential forms. Mathemati- cians invented forms at the beginning of the century and physicists working in general relativity have been using them for some twenty years. Forms give one a certain ease in calculations, elegance of notation and, usually, a more ‘generalized viewpoint. The theory of differential forms is partially motivated by the consideration of how integrands of multiple integrals 1139 Am.J. Phys. $1 (12), December 1983 ‘transform under coordinate transformations. Consider the integral ofthe scalar function f(x,y) over a two-dimension- al region R:5,f( x,y}dx dy. Under the transformation to polar coordinates, x = rcos 8, y = r sin 8, the integral be- comes Jp fir-cos 0, rsin 6) rdrd@. That is, the function Foxy) ‘transforms’ as a scalar field and the term ‘dx dy--r dr dé (or the Jacobian multiplied by dr d0 ). One does not calculate dx and dy as functions of dr and dé and ‘then multiply them. The point is that the integral and dx dy exist as a whole and hence must be transformed as such. It is conceivable though that one could construct an algebra of quantities like dx and dy that, among other things, would transform appropriately under coordinate transfor- mations. The theory of differential forms accomplishes this. In the course of studying differential forms, 1 came across the Hodge decomposition theorem which states: Ev- ery p-form can be decomposed into the sum ofthe (exterior) derivative ofa (p-1)-form, the coderivative of a(p + 1)-form and a harmonic term, and, moreover, the decomposition is unique. I realized this to be the differential form analog of the well-known Helmholtz. theorem, used in electrody- namics, which states: A vector field can be uniquely de- composed into the gradient of some scalar field, the curl of some vector field and a vector that satisfies Laplace's equa- tion. Tt became apparent that one could translate any state- ment about vectors into a statement about forms. In fact, as ‘mathematicians well know, the entire classical vector anal- ysis is subsumed by the theory of differential forms as ap- Plied to a three-dimensional space. One can derive all the theorems in vector analysis from theorems about forms in © 1983 American Association of Physics Teachers 1139,addition to obtaining all the standard vector identities— most of them quite effortlessly. I intend to use this latter fact as a vehicle to introduce the theory of differential forms, by first outlining the general theory, then by show- ing the operation on forms that corresponds to each oper- ation on vectors, and then proving many of the standard vector identities. Section II will be an unrigorous but rather intuitive ac- count of the theory of differential forms. This will then be followed by (Sec. IIT the application of forms to a three- dimensional space and the resulting connection to classical vector analysis will be made. The standard vector identities will be proved and theorems about vectors will be derived from theorems about forms {not all of which will be proved). I will then conclude with a brief section (Sec. IV) ‘on how Maxwell's equations can be phrased in the lan- guage of forms. IL, THEORY OF DIFFERENTIAL FORMS ‘A. Exterior algebra: The algebra of forms! I will be working in an n-dimensional Euclidean space with rectangular coordinates x!,x’,.,x", a point being de- seribed by x= (247 ).0") ‘At each point x we consider objects dx! /=1,2,.0m, called our basis one-forms. An arbitrary one-form a, will be a linear combination of basis one-forms: o = 37,4, dx’ Thats, the set of one-forms at a point form a vector space.” The sum of two one-forms w, = Ea, dx’ and o, = 3b, dx’ is given by o, + @2 = Bla, + b,)dx. ‘One can construct new objects called 2:forms by defining ‘a product of basis one-forms, called the exterior or wedge product. Thus a typical basis 2-form will be dx! dx! with ‘A denoting the wedge product. Once again, an arbitrary 2- form will be given by a linear combination of basis 2-forms. The wedge product is to satisfy the following rules: (a) dx Ndx' = — dx! Ady, ie, the product is anticommuta- tive; (b} (2a, dx’) Ady’ = Ba,(dx’ Ad! ie. the product is distributive. Note that in particular, (a) says that dx! Nd! for i=j. Thus, since dx'Adx — dx’ Adz’, the number of independent 2-forms at a point is given by n(n — 1)/2. (One can extend the above product to the multiplication ofan arbitrary number of one-forms to get a basis p:form: dx Adz N= Ax”, Note, however, that ifthe dimension- ality of the space is m and if p>n, then at least one of the dex‘'s is duplicated in this product. By anticommuting the necessary number of times we will obtain in the product a term of the form dx’ A dx'— which is zero. Hence ap-form is zero for p>, and we need only consider p-forms for psn. ‘Note again that an arbitrary p-form w will be given by @=8,,_,@,,,,dx"A~ Ade”. By using the same argument as I did for 2-forms, itis obvious that the number of inde- pendent basis p-forms is n!/plin — p}. For example, ina three-dimensional space, the basis one- forms at a point would be dx', dx", and dx’. The basis 2- forms would be dx! A dx", dx! Adx', and dx? Adz”. There is only one basis 3-form: dx' Adx? Adx*. Thus arbitrary one-,two-, and three-forms would be written, respectively, as 1140 Am. J. Phys, Vol. $1, No. 12, December 1983 =a, dx! +a, dx? + a,dx? = Ea, dx’, @ = ay dx! dx? + ay dx! Nd? +, dx° Nd = Ya, dx! Ady’, = ads! Ads Ade’ Now all of our considerations were at a point of the un- derlying space x!, x°,..,x". can extend these ideas to a field of forms by defining forms at each point such that they ‘are smoothly connected. Thus a field of one-forms would be given by «= Za,{ x}dx' where a,( x) is a set of differen- tiable functions rather than set of numbers. This of course defines a one-form at any given point x=x, by ( Xo) = Ea, (xoldx! This is generalized to a field of p-forms in the obvious way, and then one has for an arbitrary p-form, = Sa, (aids Adx* Au Adx’. A zero-form is defined as a scalar field. T now introduce a key idea that turns out to be at the heart of the possiblity of introducing a vector calculus. ‘This is the Hodge star operator, or dual, which associates to every p-form o an (n — p)-form *e (this dual is not to be confused with the idea of dual vector space). ‘The dual of a basis p-form dx A dx" A Adx'*is defined to be the (n—phform €,.,,,,,.,dx”'' Aw» Adx%, where the Set [igysiyap ys ody] = [integers from 1 to m}, ie, the i, jai, “complete” the Set ij.niy Taking again a three-dimensional space as an example, one has tds! = de Ade, td? = de Ndx!, Md =dx' Ade. Also Max! Ads") = dx’, *dx! Ads?) Max? Adx'|= dx", "fa fds! Nd Ade’, “dx! Ndx? Ads!) = 1 The dual of an arbitrary p-form is then defined to be the sum of the duals of its component elements, ie., the dual is a linear operator. Note that in 3-space orie has *(*0) = o. ‘As we will see later, the dual will allow us (in 3-space) to identify a pseudovector with a vector, thus allowing for operations such as vector cross product and curl Before closing this section I would like to mention an important identity that forms satisfy: «A =(—P'n No, where is ap-form and 7 isag-form. Thisis obvious for oa basis p-form and 7 a basis g-form, since each of the g one- forms of 77 must be commuted through each of the p one- forms of thusintroducing a factor of ( — 7°. The result for abitrary forms follows by distributivity. The followings a summary of the algebraic properties of forms: (1): A general p-form can be written as Dai, (ade An Ad —dx, AQ} Wa = Za,,.,, dx! A Adx? and = Bb, ,, dx’ Aw dx then +9 = Bla, +6,,,)d* Anode If isa p-form and ag-form, we have Nathan Scheer 1140AB): ON =D Yi, By yd Nem Nd dx Am Ade, ssociativity and distributivity, and ohq=(—Py ho. At B. Exterior derivative The exterior derivative de of the p-form « is defined to be the (p + 1)-form, a, ; \ do = Yodel nde! Aw Ndr Note that if is an n-form then da» = 0. We will see later that the d operator embodies in it (for n= 3) the gradient, curl, and divergence operators, de- pending on whether « is a zero-, one-, or two-form, To get ahint of this, let us examine what do looks like for various w. [fw is a zero-form, i.,a scalar function f, then =S Law, ‘which is reminiscent of the gradient operator. If is a one- form, a = $ = 3a, dx’ then dey = 6a, db = dwndx, Lae and since dx! Ad’ is skew-symmetric, = 15 (%-Boenae Bay _ oe Jax? A dx’, * ve a * Which is reminiscent of the curl operator. The precise con- nections will be made in Sec. IIT. The following useful lemmas can easily be proved": L(I: d(pAq) = dp Ag + ( — ¥pAdg. This applies to arbi- trary p-forms and g-forms. In particular, if fandg are func- tions, and ¢ and are one-forms, we have (ald (fe) =dfa + fag, (b) db) =dfNd+fdd, (hd GAd)=dbAd—dAdy. 2): If 6= Sha! 1 & dy = 8, dv = ax, azo py Thus the one-form dx! can be thought of as the exterior derivative of the coordinate function x’. With this interpre- tation we can see how forms will transform if we undertake coordinate transformation on the underlying space. Par- ticularly if," Hol y.o5F"), or F = f(x) i., we can think of the new coordinates as functions defined on the space, then x o ax Applying L(2), we then have dz! Adz? Adz? = (Jacobian) dx! Ndx" Ndx°. This explains the transformation proper- ties of integrands. I return to the example given in Sec. I of Jf xykdx dy. When integration is properly formulated in terms of forms, this would actually read 5 f(xyldx Ady. ‘Then under the coordinate transformation X=rc0s 6, y= rsin 6: Sl xydx Ady = fir cos 8, rsin 8)J( x70 \dr Nd8, by L(2}. Thus if one wants to treat dx and dy as algebraic, objects one can do so provided the algebra of forms is used. In addition to the exterior derivative d, one can define the coderivative 5, by 50 = *(d *o). The coderivative thus maps a p-form into a (p—1)form. The combination 5 + 6d is defined as the generalized Laplacian operator, since for n= 3 it is precisely that classical operator (see Appendix A). Ff. ay. C. Important theorems The three fundamental theorems of differential forms: T(1): Poincare lemma. If @ is an arbitrary p-form, then dq =0. The proof is essentially trivial: = Day, dx" N~- Ad’, do as oe ae dx! Ndx' Nm Ads, cA a Po Pies Fare OF Ade Nd Now Ade Pa, = SS de nd nds! Aw nae? by interchanging dummy indices j and k, and ths is equal 0 Feit deh Na Ada! Av Ad! “a be! Aw Nd ~d?w =0. by the skew-symmetry of dx’ Ade“, and the equality of mixed partial derivatives. This will prove to be a very powerful theorem. (2): Converse of Poincare lemma. If di =0, where Aisa p-form, then there exists a (p—I)-form @ such that 4 = do. This has certain restrictions on the topological na- ture of the domain of its validitiy, and I will not state its proof here. T[3): Hodge decomposition theorem. Any p-form can be Nathan Schleifer 1141written as@ = da + 06 + y, wherecisa(p— 1}-form, Ba (p+ I}form, and 7 a harmonic p-form, ic, satisfying Laplace's equation 4y = 0. The forms da, 58, and y are unique. Finally, one can prove, by defining integration appropri- ately (which we do not do in this paper—see, however, the comment on integration in Sec. IB, and also Appendix Di ‘T(4): Generalized Stokes theorem. If« is ap-form and Da (p+ I}-dimensional domain, then Spo = pda. that is, the integral of @ over the boundary of the region Dis equal to the integral of the exterior derivative over the region D. In three-space, when a is, respectively, a one- and two- form, we will see that this theorem encompasses the duo of, theorems, Stokes and Gauss, Moreover, for @ a zero-form we get the fundamental theorem of calculus. IL. VECTOR ANALYSIS A. Vector-form correspondences The key idea is that one can associate to every vector a cone-form and to every operation on vectorsan operation on forms. Since I will be doing vector analysis in a three-dimen- sional space, I restrict the space on which I have defined form tobe n= 3. The basic correspondence that one makes is that to every covariant vector field A = A, 8; + dat, + Asts, one asso- ciates the one-form $, =A), dx! + 4s dx? + Ay dx, Sym- bolically, I write A= AR + Asks + Asie be = Ay di! + Ay de? + Ay ds? 4, will always represent the one-form associated with the vector A.! The symbol «+ denotes the association of a vec- tor to an appropriate form. To the function f one associates the 0-form f. Tnow show how to obtain the correspondences between operations on vectors and operation on forms. 1. AXB= (AaB) — AsB}R, + (AsB, — ABslts + (ABs — ABs T would like to know what operation on ¢, and ¢p will yield a form ¢,.,.y, i., the form which is associated to the vector AX. Note that 4 Ab_ = (A, dx! +A,dx?+A,dx) AB, dx! + By dx* + By dx") = (A,B, — A,By\dx' Ndx? + (A.B, +B, — A,BsMdx? Ads". The coefficients of the basis forms look right, but this is a two-form, so how do I associate a vector toit? Simple, I first take its dual—which gives me a one-form—and then asso- ciate a vector to this one-form. So I claim that AXB=C--*16, Abs) = 8c Pictorally, I write AXB ret "bs Abn) = be To repeat, I can mimic the cross product between vec- ABylde hd? 1142 Am.J. Phys, Vol. 51, No. 12, December 1983, tors, by taking the dual of the exterior product of the corre- ‘sponding forms. Thus to find the vector given by the cross product of the vectors A and B, I write down the associated one-forms 4, and ¢p, take the dual of their exterior pro- duct, which turns out to be a one-form, and then find the vector associated to this one-form. 2.AB= A,B, + ApBy + AB, Since AB is a scalar, something must be done with 44 and ¢, as to end up with a O-form. It must be an algebraic operation since there are no derivatives involved. The only® possibility is*(d, A*4q). Note that $, isa one-form, bp one-form, so *by is a 2-form.6, A*%, a 3-form, and *(b, A%4q) a O-form. So we guess: AB*d, A*4p). It turns out that by using this type of argument, one can get all the vector operations. To check our conjecture: "1b, N*bp) = *1(A, dx! + Ay dx? + Ay dx’) A*(B, dx! + B, dx? + By dx") = *[(A, dx! + A, dx? + Ay dx’) AiByde Nae + Bde Ade! + Bydx! Ade] = *UAB, + ABs + AyB,) dx! Ads? Ads] = A,B, + A;B; + AsBy. One can check that *(6_ A*$,)= "lb, A%Gq), as it in- deed should be since B-A = A-B. now demonstrate this procedure for a vector operation involving differentiation, and then merely list all the other correspondences, 2B, aB,). (2: 2s) a. vxp= (Ms %),, (Fi), 0-(2- BSF 2B, 8B,). (Se) This clearly looks like dy —which is a 2-form (see Sec. IL B}-sowe take its dual and guess that VX B+>*(ddp). TO check doa) Id (B, dx! + B, dx? +B, dx*)} OB, oB, af Braet nde’ 4 Pade naw" [( Seperate + Shae na’) + (Ba ‘nds? + Beds rds’) ox! ax Bae nae + Braenae)] 2B 4 Bris nd + (Ghee nae + 3 x’) (2B Ba yes , (9By BB geo -(B- BY +(e He} a8, aB,\ 5 +(Ga- ae which clearly 0B. ‘T now give a complete list of the various correspon- ences Cll: A= ALR, + Aske + Asters de! + Aa +Asdx =, Ci2): fas a scalar function) +/ (as a 0-form). Nathan Schleifer 1142): Bold, Ad, VBe+*[d(*bp)]- V food. AXBer*ib, Ada). VXBeiddy). ‘Thed’s are all one-forms, allowing one to unambiguous- ly make the distinction between C(4) and C(7}, and C(3) and, 6). *(bn \*b,). B. Vector identities I list the various identities and the corresponding state- ments in the language of forms. Al: We) =e + 8Vfod (fe) =S dg + 8 df 12}: VAFA) = (VFA) + F(V-A)* [4 *764)] = *dfA%b,)+f%(d *b,). 13): YAXB) = BUVXA)— A(VXBho*[d* 16, Ads] [42 Add] —*{b4Ndde]- Tid): VX(FA) = (WAKA +S/IVX Ale [4 (79 4)] *dfNd,) + fdb,). 1(5): A(BXC] = C{AXB) = BCX A}-+"16, Nby Abc) = "bc Nb4 Nb) = "ba Nbc AGa) AX(BXO) HAC) — CIAB)e>*[d, A*lbp Ndc)] =b0*[b.N*bc] — bc*[d4 A *bn]- 1(7): (AX BCX D) = (A-CB-D} — (A-D\B-C}e> "(bs Abe) Nibc Nbn)] 1b. NOcIbn N*bn) 1b. N*bn)*\b2 A *8c). All of these correspondences of identities are obtained by using C{1}-C)7}, The proofs of the form identities are mostly straightfor- ward.’ The first four identities are all contained in L(t), which states that d (pq) = dp Aq + ( — ¥p Ada. U5) fol- lows from A(4}. The proofs of I(6) and I{7) are contained in Appendices B and C (they are the only ones that are not totally trivial. 18}: VX{V g) =0-+%d 77) =0. 1(9): VAVXA) = Oe+*(d °$,.) = 0. These two form identities follow simply from the Poin- care lemma T(1). The only distinction between the two is ‘that of applying the lemma to a 0-form and to a one-form. One thus sees that many vector identities Follow from the same form identity, only applied to forms of different o- ders. This is in fact characteristic of most all vector theo- rems and identities—the distinction between various vec- tor statements is one only of degree in terms of forms. C. Vector theorems ‘The following two theorems follow from the converse of the Poincare lemma: VII): VXE=0= 3 scalar ¥ such that E=Vy with @ a one-form => 3 a O-form @ VT(2: V-B=0=3 a vector A such that B=VXA 1143 Am.J. Phys, Vol. $1, No. 12, December 1983 <+4(d*e) =0= 3 a oneform 8 such that *o = 8 or w= *(dB), The first is simply the converse as applied to a one-form and the second as applied to a two-form. Similarly, using the Hodge decomposition theorem T\3), Teet VT(3}: Forany vector field A, 3 /;B, and Q such that arbi- trary A= Vf + VXB + Q, where V°Q = 0 «+ For any p- form «, 3 a, B, and y such that o = dar + 68 + 7 ( y har- monic). T thus obtain the Helmholtz theorem. The above was obtained with « being a one-form. If I choose w to be a 2- form, an identical theorem is obtained. Finally, making use of the generalized Stokes theorem Tih VT(4): Gauss: fre v= fea dnfarise) fits VT(5): Stokes: firxman = faa Ie +f abe = fe The first of the above is obtained by choosing in TI4), @ = *$¢,2-form, and the second by choosing = by, & cone-form. See Appendix D for the exact statement as to how these relationships are obtained. IV. ELECTRODYNAMICS For completeness, I conclude with a brief description of how Maxwell's equations are expressed in the language of forms." Since the appropriate underlying space for electrody- namics is Minkowski space—which is pseudo-Euclidean rather than Euclidean—one needs to generalize the pre- viously developed concept of duality to a space with arbi- trary metric (the previous sections assumed a metric that was specifically Euclidean). In addition, one needs to ex- tend the association between vectors and 1-forms to an as- sociation between higher-rank tensors and forms. The more general definition of dual is given in terms of the inner product defined on the space.” The definition I use, however, will be more restrictive, though sufficient for our purposes. ‘The dual ofa 1-form = J, dx" is defined as the 3-form #3 = Exam" dx" Ndx? Ndx", and the dual of the 2-form F=F,,dx'\dx" is, defined as the 2-form B= bean, Fh dx de? ‘As previously, one associates to the current four-vector J, the L-form J,, dx“. To the covariant skew-symmetric electromagnetic field tensor F,,, one associates the 2-form LF, dx" dx", i.e.,F,. (tensor) «AF, dx” \dx" (2-form). ‘Maxwell's equations can then be shown to be given by vB=0 XE + LSB gf Fa) =OedF = © VE =4np 1 aE _ de jJore, = tye yxp—49F _ An oF, =") cae od P= ATs, where [1 v(2 ] denotes skew-symmetry in all the indices, Nathan Schleifer 1143and a comma denotes differentiation with respect to the index following it. The expressions above are, respectively, Maxwell's equations in vector, tensor, and form notation. The four vector equations thus get reduced to two sim- ple-looking differential-form equations dF=0 and d*F=(4n/c}*J. As usual, the simplicity of notation has been arrived at by constructing a deeper strata of definitions. ‘One can similarly express all the relevant electromagnet ic quantities, -., the Lorentz force, Poynting vector, field energy, etc. in the language of forms! It is within this language that the question of the apparent nonexistence of magnetic monopoles is most compelling.* V. CONCLUSION Thave shown how classical vector analysis is not only ‘completely contained in the theory of differential forms, but also how the latter conceptually unifies many appar- ently disparate vector ideas. From the point of view of differential forms, it is the conjunction of the concept of dual and the particular di- ‘mension of three that allows for a vector calculus. This can be seen in the following manner. tis assumed that a vector is identified with a form of a given order. The dual of a form, providing an identification of an (n — p)-form with ap-form, allows oran introduction of differentiation which changes the order of the form, by identifying the higher-order form with a lower one, thus ‘remaining in the space of forms identified with vectors."" If the dimension of the space is larger than three the space will not be closed, in the sense that the operation of exterior product or differentiation may take one out of the space of forms that can be identified with vectors. Pictorial- ly, one has. O-form —\-form 2-form 3:form 1 1 1 L sealar —weetor pseudovector pseudoscalar Inany higher-dimensional space this would not be possi- ble, as there would be forms that cannot be associated with vectors. ACKNOWLEDGMENTS My thanks are extended to Bronek Czarnocha for many inspiring discussion as well as a critical reading of the manuscript, and to Marlene Stulbach for her spiritual sup- port. APPENDIX A Proof that dé +- 8d agrees with classical Laplacian oper- ator when applied to a function f: 1144 Am. J. Phys, Vol 51, No. 12, December 1983 (a5 + Sd\f =(d*d* 4 *d*ayp =*d*f= sae(y Lax) “a(S en Lae nas’) -(5 ests nds*) Semen ge t ax ax af 3 a igre Zar ‘where the second equality arises from the fact that since *f=fdx' Nd? Nx! then d* f= 0. qed. APPENDIX B Proof that * [4, "Ip Adel] = 45 “lb, A*Bc} — 6.91b, A*b a): "by Adel = (oe. ax") a(Se,ae | =+(ga.6, dn nee’) = yp, C6, dx Then b.NMbe Abc) = (3. a)a5a, Cty a2 Since anne quit must be ehe jo. Thus b4N*1ba Abc) =D [1d,, dx" +A, dx") NB,Cae = A,B,C, Eiuude Ade” + 5A, AS dx, C6 Then *[b. Abe Gel] wri OE = TAB Cabin YA Bi, C6 inns EE A,B,C, dx? + A,B,C, dx! —b Abs A*bn) + bab, N*bc) ged. This immediately translates into the vector theorem AX(BXO) = BIAC) — CAB). APPENDIX C Proof of (AX B}(CXD) = (A-C\(B-D) + (A-DIB-C) In terms of forms we must prove that [Nathan Schleifer 1144"(bs Ada) Abc NGo)] = "bs Nb cI"ba N*b0) — ba N*b "lbs A *Oc). From the previous proof, we have Hb Mbp) = Dei By, dx and (6cA$) = DCD, ax" Ndx Then "bs Mal Nbc NOo) =D 6 yA B,C, D, dx dx” Adz*, Clearly 7 i,#i, and ii, Ai, so we must have either OF iy = iy, iy = ty, Then iba Adal Albc Abo) = TA, BC, Dy ens de Ads" Ade + A, B.C,D, 6,45 dx! Ade" Nas and *[4, Nb) Nib Ado) = TA,B,C, Dy Eins +3A,B.C.Die = 24, B,C,D, ZA BCP “ibs A*bel*ibe A *bp) bs Abn) ibe A *dc) APPENDIX D ‘Schematic derivation of Gauss’s and Stokes’s theorem from the generalized Stokes’s theorem T(4): (Gaus: { vvar= [va fe knve the falling eomapadenecs of vector 10 forms: fener er BW , BS Ox, ae, * Oxy mtr = (Seg Bey Ox, | Ox, ax var =( aso? ds? x! Adxt Ade and . Ved A= Vids! di? + V,de dx'+ Vide dv sotby = Vide Adv+ V, de Adx! + Vy dx! Ads? Then, by choosing o = *4,, a form, we get Jpvar=[vdan [aver = [re (ii) Stokes [yxaas 5 ard Ds Le dS = dx? dS, + ds? de! 8, + dx! dx? 8, 1145 _Am.J. Phys, Vol. $1, No. 12, December 1983 then vicads = (24 — aie as? ox! Ox? (4-4) Jas! de ax Ox aA, _ aA, (TAs _ AAs 5s 2 +(Se-H) db, (@ - Vis nae GA, AsV ies nae! (BAL _ 9s \ ip na +(S-3) Ay AA yes pages 2 As Vast Ade. +(G-Se) Similarly, A-dlesd,. Then, by choosing o =,, a one form: frxaas Gardter a, = [be "Present address: Department of Physic, Seton Hall University, South ‘Orange, New Jersey 07079 "A briefcomment on notation: For referential convenience I abel various Sets of equations with alphabetic letters, TIS), that have the follow: ing meanings: A—Alebrac properties of forms, L—Lemmas pertain ing to forms, T—Theorems pertaining to forms, C—Correspondences between vector operations and form operations, correspondences between form Identities and vector identities, and VT—Vector Theorem correspondences to form theorems. J(xy77.8) isthe Jacobian of the ‘transformation from variable 0 to variables xy. The symbols and ¥ generally represent one forms, a distinct from « whose order is arbi- trary. The letters pand g, though usually denoting the order ofa form, will occasionally represent a form itself ofthat order. The limits of summation willbe omitted when itis obvious, The space dual othe tangent space a the pont. See, for example, H. Flanders, Diferential Forms (Academic, New ‘York, 1963) “Therelevance of this to fermion creation and anniilation operators be presented ina Tutute paper. “This association is easily sten tobe preserved under a coordinate trans- formation, ‘As wel asthe symmetic product“, A%4,|-See below. “I would like to mention thatthe only typeof vector identity Ihave not been able to proveis,e, VXIAXB) = (BA ~ (A-VIB 4 (BIA — (FAB, believe thet thisis because it involves an abuse of vector notation, ¢, ina term like (BVIA. I is conceivable, though, that a corresponding abuse ofthe form notation would yield the differential form analog of this typeof vector identity “This topic, as wel asthe general theory of differential forms ina four- dimensional Minkowski spac, will be discusted in detail n a fortheom- ing paper. Note that I adopt the Einstein implied) summation conven- tion inthis ection “An even more general definition can be given in terms of the “volume n= form.” See B. Schutz, Geometrical Methods of Mathematical Physict (Cambridge University, Cambridge, 1980), Chap 4. Ths isan excellent book on mathematical physics that san elegant compromise between 8 physicis's desire for simplicity and a mathematicians for rigor. "See C. Misner,K. Thorne, andJ.A. Wheeler, Gravitation Freeman, San Francisco, 1973, Chaps. 4 and 8 "Which, in the language of vectors, is the identification of pscudovector and a vector, allowing for the introduction of curl and erors produc hile sil remaining in the space of vectors, X Nathan Schleifer 1145